Study of Region Convolutional Neural Network Deep Learning for Fire Accident Detection

Fires accident is one of the disasters which take human life, infrastructure destruction due to its violence or to the delay for the rescue. Object detection is one of the popular topics in recent years, which can play the robust impact for detecting fire and more efficient to provide information to this disaster. However, this study presents the fire detection processed using region convolution neural network. We will train images of different objects in fire using ground truth labeling. After labeling images and determining the region of interest (ROI), the features are extracted from training data, and the detector will be trained and will work to each and image of fire. To validate the effectiveness of this system the algorithm demonstrates images taken from our dataset.